i had 1536 and t5 mask checked... i unchecked t5 and left the 1536, running using 21.2GB vram but ge
i had 1536 and t5 mask checked... i unchecked t5 and left the 1536, running using 21.2GB vram but getting about 26s/it 
Ngetting this after trying to start. Issue with dir name fixed, new ones come))
Ntried to start rank_5
Fuse dev fp16 model
NI'm using this one
Fyes it is correct one
Fwhat is your gpu model?
Nmaybe t5 is wrong?
Fyep it may be wrong :d
Nrtx 4060ti 16gb
Fuse the downloader bat file i have
Nhad to extract only t5 from your python, because downloading flux again will piss me off)
Nthx
Fyou can edit my files
Fremove the ones you dont need
Fall my files are editable easy
Nyep, just did that, thanks
X
Xjust put
kohya_ss\wd14_tagger_model
Xnvm, downloaded another model from kohya and replaced that model with eva02, it works
X
EHas anyone received this issue while trying to traing flux with Kohya?
Used the recent single click installer V15 from SECourses Patreon.
Used the recent single click installer V15 from SECourses Patreon.
EI know it is probably something to do with accelerate
Fwhich base model
Fyou given?
Ethe base model form the Blacklabs repo on hugging face
Efrom*
Ffile name and size please
FYouTubeSECourses
Many people having issues due to inaccurate Accelerate setup. This is how to setup to make it work properly
1-click installers and best configs for Kohya FLUX training :
A Big complete tutorial very soon hopefully : https://www.patreon.com/posts/kohya-flux-lora-110293257
1-click installers and best configs for Kohya FLUX training :
A Big complete tutorial very soon hopefully : https://www.patreon.com/posts/kohya-flux-lora-110293257
Eflux1-dev.safetensors
Fok that is accurate. what is your T5 model name and size?
EI have tried the combined version: t5xxl_fp16.safetensors
and the non combined version from the blacklabs repo:
and the non combined version from the blacklabs repo:
Fplease also apply accelerator setup above and restart and try again
E
Fthis is wrong
Fyou need to use this one
Fcomfyanonymous/flux_text_encoders
Ekk will grab and retest
Fyou know i have model downloader too 
EExtra note I do have a 4090 and a 3080 plugged in to the same rig. Though I usually only use 4090 for training 3080 is mainly there for other uses. So I normally set accelerator to only use GPU ID 0
Fi have rtx 3090 and 3060
Fshouldnt be a problem
Edo you have both enabled in accelerator ? I thought it would cause the 3060 vram to max out and crash the training?
Fnope it uses only single gpu
Fduring training
Funless you set multiple gpus
EKk so what I though so weird
